Subscribe to RSS
Objective hearing threshold estimation based on ABR measurements
18 April 2018 (online)
A standard procedure for hearing threshold determination in neonates is based on the visual evaluation of ABR responses by an experienced clinician. Unfortunately, this procedure is highly error-prone as threshold estimates of different evaluators may differ by up to 60 dB (Vidler and Parker, 2004, Int J Audiol. 43:417 – 29). Here we present a fully objective method for the evaluation of ABR data to overcome these limitations.
Our approach was validated using extensive numerical simulations and ABR data from rodent models.
ABR based hearing threshold are defined as smallest sound intensity leading to an evoked neuronal brainstem response. For threshold estimation from physiological responses it is common practice to use stimulus intensities close to the putative threshold. Trivially, the signal-to-noise ratio (S/N) is worst near the threshold since the intensities of evoked responses are positively correlated with stimulus strength. In other words, thresholds are usually determined from measurements with low S/N, and consequently prone to errors. We demonstrate that the threshold can objectively be estimated by fitting of a generalized hard sigmoid function to the ABR amplitudes. The offset of this hard sigmoid function is fixed at the amplitude of the background-noise leading to an increased robustness of the fitting approach and allowing for the determination of a hearing threshold without near threshold measurements nor arbitrary definition of some threshold criteria.
In conclusion, the novel fitting approach provides the opportunity to fully reproducible determine a hearing threshold without defining a criterion based on an arbitrary parameter and provides robustness against a reduction of measured supporting points as well as the number of measurement repetitions.